Research On Modeling Methods For Cross Piercing Energy Consumption And Piercing Process Of Seamless Steel Tube

With the development of modern, the applying field of seamless steel tube is enlarging and requirement is becoming much higher. Now countries put more emphasis on energy conservation and emission reduction for lower production costs. In this case, studying the cross piercing energy consumption of seamless steel tube is very necessary. The cross piercing process as the first deformation process in the production of seamless steel tube plays an important role, and the piercing energy consumption as an important indicator of piercing process also is very important for the whole production. Because the piercing process is very complex and has multi-period、nonlinear etc, The relationship between the piercing energy consumption and process variables is very difficult to be determined by mechanism model.In this thesis, using the methods of soft measurement modeling build the correlation model between the piercing energy consumption and process variables, and this model is updated online. At the same time, the thesis establishes the model for the cross piercing process of seamless steel tube to simulate the actual production. This model can be used to verify the effect of parameters.On the background of the cross piercing process of seamless steel tube of Baosteel, this thesis mainly researches on mdeling methods for the piercing energy consumption and the piercing process. The main content for this thesis is shown as follows.(1) This thesis takes multi-way partial least square (MPLS) algorithm as the core for modeling of piercing energy consumption of seamless steel tube and makes some improvements on it. In MPLS algorithm, using PLS builds the model for two-dimensional data, but PLS is suited to local nonlinearity. We need use nonlinear PLS to model because of the complicated nonlinear of piercing process. This thesis retains the external model of PLS and uses the extreme learn machine (ELM) to nonlinear fitting for the internal model of PLS. Considering the unstable output of ELM, this thesis uses the self-adaptive threshold AdaBoost.RT to overcome this instability and improve the forecasting precision of model. The method of nonlinearMPLS modeling based on adaptive threshold AdaBoost. RT has been applied to building model for the piercing energy consumption and Makes a good forecasting precision.(2) In order to fit for the change of working condition, it is need to update the model online. Based on the MPLS algorithm, online sequential ELM can be used to update the internal model of PLS while weighting the loading matrix for the external model updating. In the end, using AdaBoost.RT to integrate the whole algorithm overcomes the instability. According to the result of simulation, the updating method can update the model in time when conditions changed. So the model can forcast the value of perforation energy consumption.(3) The thesis has established the Deform finite element model for the cross piercing process of seamless steel tube, which has some practical value. This model is not only to test the parameters obtained from soft measurement modeling, but also to provide model for the later research on hybrid modeling.